Real-Time GPU-Aided Lung Tumor Tracking

A real time solution of tracking daily lung tumor motion is proposed in this paper in order to achieve an accurate dose delivery for radiation therapy as recently developed cone-beam computed tomography (CBCT) technique is not able to catch tumor motions due to the patient´s respiration. We develop a novel GPU-based fast digitally reconstructed radiograph (DRR) generation algorithm which enables an instant DRR computation and rendering from patients´ radiation therapy treatment planning CTs. In the meantime, classic image correlation algorithm is extended as the main method to locate tumors in X-ray 2D projections, the raw data of CBCT scans. With the GPU-aided implementation, this algorithm is capable of capturing movement of lung tumors as fast as the CBCT image acquisition in real time, which greatly facilitates the radioactive treatment.